src/statichuff/statichuff.c

   1
   2 #include "statichuff.h"
   3 #include <stdlib.h>
   4
   5 void putbit(char bit, char restart, char flush, VFILE *fp)
   6 {
   7         static unsigned long int bits_buffer = 0;
   8         static unsigned char bits_used = 0;
   9
  10         /* me obligan a emitir el output */
  11         if ((flush == 1) && (bits_used > 0)) {
  12                 bits_buffer = bits_buffer << ((sizeof(unsigned long int)*8) - bits_used);
  13                 vfwrite(&bits_buffer,sizeof(unsigned long int),1,fp);
  14                 bits_buffer = 0;
  15                 bits_used = 0;
  16                 return;
  17         }
  18         /* me indican que comienza un nuevo output */
  19         if (restart) {
  20                 bits_buffer = 0;
  21                 bits_used = 0;
  22         }
  23         /* inserto el bit en el buffer */
  24         bits_buffer = bits_buffer << 1;
  25         bits_buffer |= bit;
  26         bits_used++;
  27
  28         /* lleno el buffer, escribo */
  29         if (bits_used == 32) {
  30                 vfwrite(&bits_buffer,sizeof(unsigned long int),1,fp);
  31                 bits_buffer = 0;
  32                 bits_used = 0;
  33         }
  34         return;
  35 }
  36
  37 void shuff_cpynode(SHUFFNODE *node1, SHUFFNODE *node2)
  38 {
  39         node1->symbol = node2->symbol;
  40         node1->freq = node2->freq;
  41         node1->lchild = node2->lchild;
  42         node1->rchild = node2->rchild;
  43 }
  44
  45 int shuff_compnode(const void *node1, const void *node2)
  46 {
  47         if (((SHUFFNODE*)node1)->freq < ((SHUFFNODE*)node2)->freq) return 1;
  48         if (((SHUFFNODE*)node1)->freq > ((SHUFFNODE*)node2)->freq) return -1;
  49         return 0;
  50 }
  51
  52 int shuff_rescalefreq(t_freq *freqtable)
  53 {
  54         int i;
  55         t_freq totalfreq = 0;
  56
  57         /* Divido por la mitad las frecuencias, asegurando de no perder */
  58         for (i = 0; i < 256; i++) {
  59                 freqtable[i] = (freqtable[i] >> 2) | 1;
  60                 totalfreq += freqtable[i];
  61         }
  62
  63         return totalfreq;
  64 }
  65
  66 int shuff_scanfreq_chunk(HUFF_STATE *chunkshuff, char* chunk, int chunksize)
  67 {
  68         /* Locals */
  69         int i = 0;
  70         unsigned char symbol = 0;
  71
  72         /* Contamos las frecuencias del chunk*/
  73         for (i = 0; i < chunksize; ++i) {
  74                 symbol = chunk[i];
  75                 chunkshuff->freqtable[symbol] += 1;
  76                 chunkshuff->sumfreq += 1;
  77
  78                 /* Si llegue al tope de freq acumulada, halve em */
  79                 if (chunkshuff->sumfreq == 14930352)
  80                         chunkshuff->sumfreq = shuff_rescalefreq(chunkshuff->freqtable);
  81         }
  82
  83         return 1;
  84 }
  85
  86 int shuff_scanfreq(char *inputfile, t_freq *freqtable)
  87 {
  88         /* Locals */
  89         FILE *fp;
  90         t_freq sumfreq = 0;
  91         int i,symbol;
  92
  93         /* Inicializamos la tabla de frecuencias */
  94         for (i = 0; i < 256; ++i) freqtable[i] = 0;
  95
  96         /* Abrimos el file */
  97         if ((fp = fopen(inputfile,"r")) == NULL) return 0;
  98         while (!feof(fp)) {
  99                 /* Contamos las frecuencias */
 100                 symbol = fgetc(fp);
 101                 if (symbol == EOF) continue;
 102
 103                 freqtable[symbol]++;
 104                 ++sumfreq;
 105
 106                 /* Si llegue al tope de freq acumulada, halve em */
 107                 if (sumfreq == 14930352)
 108                         sumfreq = shuff_rescalefreq(freqtable);
 109         }
 110
 111         fclose(fp);
 112         return 1;
 113 }
 114
 115 SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
 116 {
 117         int i,j = 0,nonzero = 0;
 118         SHUFFNODE *inputlist;
 119
 120         /* Calculo cuantas frequencias > 0 hay y creo la tabla */
 121         for (i = 0; i < 256; ++i) if (freqtable[i] > 0) nonzero++;
 122         inputlist = (SHUFFNODE*)malloc(sizeof(SHUFFNODE)*nonzero);
 123
 124         /* Cargo la inputlist del huffman solo con freqs > 0 */
 125         for (i = 0; i < 256; ++i)
 126                 if (freqtable[i] > 0) {
 127                         inputlist[j].symbol = i;
 128                         inputlist[j].freq = freqtable[i];
 129                         inputlist[j].lchild = NULL;
 130                         inputlist[j].rchild = NULL;
 131                         j++;
 132                 }
 133
 134         *nonzerofreqs = nonzero;
 135         return inputlist;
 136 }
 137
 138 SHUFFNODE *shuff_buildtree(t_freq *ftable)
 139 {
 140         SHUFFNODE *lastsymbol;
 141         SHUFFNODE *node1,*node2,*root;
 142         SHUFFNODE *inputlist;
 143         int freqcount = 0;
 144
 145         /* Genero la input list en base a la cual genera el arbol */
 146         inputlist = shuff_buildlist(ftable, &freqcount);
 147         lastsymbol = inputlist+(freqcount-1);
 148
 149         while (lastsymbol > inputlist) {
 150                 /* Ordeno la lista por frecuencia descendente */
 151                 qsort(inputlist,freqcount,sizeof(SHUFFNODE),shuff_compnode);
 152                 /* Tomo los ultimos dos elementos, generando dos nodos del arbol */
 153                 node1 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
 154                 node2 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
 155                 shuff_cpynode(node1,lastsymbol-1);
 156                 shuff_cpynode(node2,lastsymbol);
 157                 lastsymbol -= 1;
 158                 /* Nodo ficticio con la suma de las probs y los ptros a childs */
 159                 lastsymbol->symbol = 256;
 160                 lastsymbol->freq = node1->freq + node2->freq;
 161                 lastsymbol->lchild = node1;
 162                 lastsymbol->rchild = node2;
 163                 --freqcount;
 164         }
 165
 166         /* Copio la raiz para poder liberar la lista sin perderla */
 167         root = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
 168         shuff_cpynode(root,lastsymbol);
 169
 170         /* Free up mem */
 171         free(inputlist);
 172
 173         /* Devuelvo el puntero a la raiz del arbol de huffman */
 174         return root;
 175 }
 176
 177 void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
 178 {
 179         int i,j;
 180         unsigned short int auxcode;
 181         unsigned char bit;
 182
 183         for (i = 0; i < 256; ++i) {
 184                 if (codetable[i].codelength > 0) {
 185                         auxcode = codetable[i].code;
 186                         printf("Symbol:%i  Freq: %li  Code:",i,freqtable[i]);
 187                         for (j = codetable[i].codelength-1; j >= 0; --j) {
 188                                 auxcode = codetable[i].code;
 189                                 auxcode = auxcode >> j;
 190                                 bit = auxcode & 1;
 191                                 printf("%i",bit);
 192                         }
 193                         printf("  Length:%i\n",codetable[i].codelength);
 194                 }
 195         }
 196 }
 197
 198 void shuff_zerocodes(SHUFFCODE *table)
 199 {
 200         int i;
 201
 202         /* Inicializo los codigos prefijos */
 203         for (i = 0; i < 256; ++i) {
 204                 table[i].code = 0;
 205                 table[i].codelength = 0;
 206         }
 207 }
 208
 209 void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
 210 {
 211         if (node->symbol < 256) {
 212                 /* Guardo el codigo en la tabla */
 213                 table[node->symbol].code = code;
 214                 table[node->symbol].codelength = level;
 215         }
 216         else {
 217                 code = code << 1;
 218                 shuff_buildcodes(table,node->lchild,level+1,code);
 219                 code |= 1;
 220                 shuff_buildcodes(table,node->rchild,level+1,code);
 221         }
 222 }
 223
 224 int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)
 225 {
 226         FILE *fpsource;
 227         VFILE *fpdest;
 228         int symbol,i;
 229         unsigned long int sourcesize;
 230         char bit;
 231         SHUFFCODE symbolcode;
 232
 233         /* Abrimos el file */
 234         if ((fpsource = fopen(shuff->sourcefile,"r")) == NULL) return 0;
 235         if ((fpdest = vfopen(shuff->targetfile,"w",shuff->volsize)) == NULL) return 0;
 236
 237         /* Guardamos el size el archivo original e inputlist como header */
 238         fseek(fpsource,0,SEEK_END);
 239         sourcesize = ftell(fpsource);
 240         vfwrite(&sourcesize,sizeof(unsigned long int),1,fpdest);
 241         vfwrite(shuff->freqtable,sizeof(t_freq),256,fpdest);
 242
 243         /* Encodeo */
 244         fseek(fpsource,0,SEEK_SET);
 245         while (!feof(fpsource)) {
 246                 /* Levanto un symbolo (byte) */
 247                 symbol = fgetc(fpsource);
 248                 if (symbol == EOF) continue;
 249
 250                 /* Cargamos el codigo y lo emitimos */
 251                 symbolcode = ctable[symbol];
 252                 for (i = symbolcode.codelength; i > 0; --i) {
 253                         bit = (symbolcode.code >> (i-1)) & 1;
 254                         putbit(bit,0,0,fpdest);
 255                 }
 256         }
 257
 258         /* Hacemos un flush de lo que haya quedado en el buffer de salida */
 259         putbit(0,0,1,fpdest);
 260         fclose(fpsource);
 261         vfclose(fpdest);
 262         return 1;
 263 }
 264
 265 int shuff_encode_file(HUFF_STATE *shuff)
 266 {
 267         /* Locals */
 268         SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256);
 269
 270         /* Veo si debo armar una freqtable o si esta preloaded */
 271         if (!shuff->preloadfreq) if (!shuff_scanfreq(shuff->sourcefile,shuff->freqtable)) return 0;
 272
 273         /* Genero el arbol de huffman */
 274         shuff->codetree = shuff_buildtree(shuff->freqtable);
 275
 276         /* Armo la tabla de codigos prefijos para el encoder */
 277         shuff_zerocodes(codetable);
 278         shuff_buildcodes(codetable,shuff->codetree,0,0);
 279         /*shuff_printcodes(codetable,shuff->freqtable);*/
 280
 281         /* Encodeo byte per byte */
 282         shuff_encode_symbols(shuff,codetable);
 283
 284         /* Free up memory baby yeah */
 285         free(codetable);
 286
 287         return 1;
 288 }
 289
 290 SHUFFNODE *shuff_decode_symbols(SHUFFNODE *entrynode, unsigned long int buffer,
 291                                                          int *bitsleft, unsigned short int *symbol)
 292 {
 293         char bit = 0;
 294
 295         /* Levanto el symbolo y si es uno valido, devuelvo */
 296         *symbol = entrynode->symbol;
 297         if (*symbol != 256) return entrynode;
 298         if (*bitsleft == 0) return entrynode;
 299
 300         /* Obtengo otro bit a procesar y me muevo en el arbol */
 301         bit = (buffer >> ((*bitsleft)-1)) & 1;
 302         --(*bitsleft);
 303         if (bit == 0) return shuff_decode_symbols(entrynode->lchild,buffer,bitsleft,symbol);
 304         else return shuff_decode_symbols(entrynode->rchild,buffer,bitsleft,symbol);
 305 }
 306
 307 int shuff_decode_file(HUFF_STATE *shuff)
 308 {
 309         SHUFFNODE *currnode;
 310         unsigned long int codebuffer;
 311         FILE *fpdest;
 312         unsigned short int decoded_symbol;
 313         int bitsleft;
 314
 315         /* Levanto cuantos bytes decodeo y la freq table */
 316         if ((fpdest = fopen(shuff->targetfile,"w")) == NULL) return 0;
 317         currnode = shuff->codetree;
 318
 319         while (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0)) {
 320
 321                 /* Leo un buffer de 32 bits */
 322                 if (vfread(&codebuffer,sizeof(unsigned long int),1,shuff->decoderfp) != 1) continue;
 323                 bitsleft = sizeof(unsigned long int) * 8;
 324
 325                 /* Proceso el buffer sacando simbolos hasta que se me agote */
 326                 while ((bitsleft > 0) && (shuff->bytesleft > 0)) {
 327                         currnode = shuff_decode_symbols(currnode,codebuffer,&bitsleft,&decoded_symbol);
 328                         /* Si obtuve un symbolo valido lo emito*/
 329                         if (decoded_symbol != 256) {
 330                                 fputc(decoded_symbol,fpdest);
 331                                 currnode = shuff->codetree;
 332                                 --(shuff->bytesleft);
 333                         }
 334                 }
 335         }
 336
 337         fclose(fpdest);
 338
 339         return 1;
 340 }
 341
 342 HUFF_STATE *shuff_init_decoder(char *inputfile, char *outputfile)
 343 {
 344         /* Locals */
 345         HUFF_STATE *shuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));
 346         shuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
 347
 348         /* Init fields */
 349         shuff->coderfp = NULL;
 350         shuff->targetfile = NULL;
 351         shuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1));
 352         strcpy(shuff->sourcefile,inputfile);
 353         if (outputfile != NULL) {
 354                 shuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
 355                 strcpy(shuff->targetfile,outputfile);
 356         }
 357
 358         /* Levanto cuantos bytes decodeo y la freq table */
 359         if ((shuff->decoderfp = vfopen(shuff->sourcefile,"r",0)) == NULL) return 0;
 360         vfread(&(shuff->bytesleft),sizeof(unsigned long int),1,shuff->decoderfp);
 361         vfread(shuff->freqtable,sizeof(unsigned long int),256,shuff->decoderfp);
 362         shuff->codetree = shuff_buildtree(shuff->freqtable);
 363
 364         return shuff;
 365 }
 366
 367 HUFF_STATE *shuff_init_encoder_byfile(char *inputfile, char *outputfile, long volsize)
 368 {
 369         /* Locals */
 370         HUFF_STATE *fshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));
 371         int i;
 372
 373         /* Inicializo la estructura para trabajar con Huff Static by File */
 374         fshuff->coderfp = NULL;
 375         fshuff->decoderfp = NULL;
 376         fshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1));
 377         fshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
 378         strcpy(fshuff->sourcefile,inputfile);
 379         strcpy(fshuff->targetfile,outputfile);
 380         fshuff->volsize = volsize;
 381         fshuff->preloadfreq = 0;
 382         fshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
 383         for (i = 0; i < 256; ++i) fshuff->freqtable[i] = 0;
 384         fshuff->sumfreq = 0;
 385         fshuff->bytesleft = 0;
 386         fshuff->codetree = NULL;
 387
 388         return fshuff;
 389 }
 390
 391 HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize)
 392 {
 393         /* Locals */
 394         HUFF_STATE *cshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));
 395         int i;
 396
 397         /* Inicializo la estructura para trabajar con Huff Static by Chunks */
 398         cshuff->decoderfp = NULL;
 399         cshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(outputfile)+2));
 400         cshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
 401         strcpy(cshuff->targetfile,outputfile);
 402         strcpy(cshuff->sourcefile,outputfile);
 403         strcat(cshuff->sourcefile,"~");
 404         cshuff->volsize = volsize;
 405         cshuff->preloadfreq = 1;
 406         cshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
 407         for (i = 0; i < 256; ++i) cshuff->freqtable[i] = 0;
 408         cshuff->sumfreq = 0;
 409         cshuff->bytesleft = 0;
 410         cshuff->codetree = NULL;
 411
 412         /* Abrimos un archivo temporal para ir tirando los chunks */
 413         if ((cshuff->coderfp = fopen(cshuff->sourcefile,"w")) == NULL) return NULL;
 414
 415         return cshuff;
 416 }
 417
 418 void shuff_deinit_encoder(HUFF_STATE *shuff)
 419 {
 420         /* Libero mallocs y cierro archivos */
 421         free(shuff->freqtable);
 422         free(shuff->sourcefile);
 423         free(shuff->targetfile);
 424         if (shuff->coderfp != NULL) fclose(shuff->coderfp);
 425         if (shuff->decoderfp != NULL) vfclose(shuff->decoderfp);
 426
 427         /* Destruyo recursivamente el arbol de codigos */
 428 }
 429
 430 void shuff_deinit_decoder(HUFF_STATE *shuff)
 431 {
 432         /* Libero mallocs y cierro archivos */
 433         free(shuff->freqtable);
 434         free(shuff->sourcefile);
 435         if (shuff->targetfile != NULL) free(shuff->targetfile);
 436         if (shuff->decoderfp != NULL) vfclose(shuff->decoderfp);
 437
 438         /* Destruyo recursivamente el arbol de codigos */
 439 }